class: center, middle, inverse, title-slide .title[ # <a href="https://github.com/mca91/kausalanalyse_slides/raw/main/ResearchQuestions/ResearchQuestions.pdf">Research Questions</a> ] .author[ ### Christoph Hanck ] .date[ ### Summer 2023 ] --- layout: true <a style="position: absolute;top:5px;left:10px;color:#004c93;" target="Overview" href="https://kaslides.netlify.app/"><svg aria-hidden="true" role="img" viewBox="0 0 576 512" style="height:15px;width:16.88px;vertical-align:-0.125em;margin-left:auto;margin-right:auto;font-size:inherit;fill:currentColor;overflow:visible;position:relative;"><path d="M575.8 255.5c0 18-15 32.1-32 32.1h-32l.7 160.2c0 2.7-.2 5.4-.5 8.1V472c0 22.1-17.9 40-40 40H456c-1.1 0-2.2 0-3.3-.1c-1.4 .1-2.8 .1-4.2 .1H416 392c-22.1 0-40-17.9-40-40V448 384c0-17.7-14.3-32-32-32H256c-17.7 0-32 14.3-32 32v64 24c0 22.1-17.9 40-40 40H160 128.1c-1.5 0-3-.1-4.5-.2c-1.2 .1-2.4 .2-3.6 .2H104c-22.1 0-40-17.9-40-40V360c0-.9 0-1.9 .1-2.8V287.6H32c-18 0-32-14-32-32.1c0-9 3-17 10-24L266.4 8c7-7 15-8 22-8s15 2 21 7L564.8 231.5c8 7 12 15 11 24z"/></svg></a> --- exclude: true <style>.xe__progress-bar__container { bottom:0; opacity: 1; position:absolute; right:0; left: 0; } .xe__progress-bar { height: 0.25em; background-color: #004c93; width: calc(var(--slide-current) / var(--slide-total) * 100%); } .remark-visible .xe__progress-bar { animation: xe__progress-bar__wipe 200ms forwards; animation-timing-function: cubic-bezier(.86,0,.07,1); } @keyframes xe__progress-bar__wipe { 0% { width: calc(var(--slide-previous) / var(--slide-total) * 100%); } 100% { width: calc(var(--slide-current) / var(--slide-total) * 100%); } }</style> # Empirical research .vcenter[ - **Empirical research** is any research that uses structured observations from the real world to attempt to answer questions. - **Quantitative empirical** research is just empirical research that uses quantitative measurements (numbers, usually). More data sets, fewer interviews. - One problem with quantitative empirical research is that the numbers that are observed often don’t tell exactly what we want to know. - It turns out that, by doing right kind of research, we can get the *right kind of numbers*. By carefully designing the right kind of analysis, we can get the answer to our research question from the numbers. ] --- exclude: true # Why research needs a design .vcenter[ - A lack of solid research design can be seen in the results, as well. - Different studies seem to give different answers to research questions because they are not actually answering that question in the first place. - For example: - Scientists may claim `\(2+2\)` only has one answer - If you’re actually calculating something entirely different from `\(2+2\)`, you might come back with an answer of `\(6\)`, or `\(1\)`, or `\(-52\)` - Then you wake up to a news headline reading that scientists have determined that `\(2+2=-52\)` - There may not be a straightforward answer to our research question, but we must continue to try! ] --- # What is a research question anyway? .vcenter[ .blockquote[ ### Key Koncept: Research question A **research question** is a question 1. ... that *can be answered* and 2. ... for which having that answer will *improve our understanding of how the world works*. A research questions takes us from a **theory** to an **hypothesis** that we can check using real world data. The data may provide **evidence** in favor or against our hypothesis. ]] --- # What is a research question anyway? &mdash; ctd. <img src="data:image/png;base64,#researchquestion.png" width="85%" style="display:block; margin-right:auto; margin-left:auto;margin-top:10%;" /> --- # Research questions .vcenter[ .blockquote[ ### Example: The influence of smartphone usage on children - Does heavy smartphone usage damage children's eyesight? - Does smartphone usage stimulate creative or intellectual activity? - Does smartphone usage reduce children's ability to concentrate? ]] --- # Why not simply look for patterns in the data? </br> - We are living in an information economy, surrounding us by an increasing amount of readily available data - Why not skip hassling with theories and research questions and simply look for *patterns* in the data? - A lot of people do &mdash; this commonly referred to as **data mining** <br> .blockquote[ ### Definition: Data Mining Data mining is the systematic application of computer-aided methods to discover patterns, trends or correlations in (often large and connected) data sets. Algorithms based on statistical methods are frequently used for this purpose. ] --- # Data mining </br> ... is a useful toolkit if we care more about the *what* rather than the *why* : - Data mining is often good at - ... making *predictions under stability*. - ... finding patterns in the data. These may foster research questions that we can examine further in other data sources. - Data mining is less good at - ... helping to come up with a good theory or to improve an existing one. - ... producing valid inference. It has a tendency to find **false positives**.¹ .footnote[[1] A false positive is a result that incorrectly indicates the presence of a condition under examination.] --- # Data mining &mdash; predictions under stability .vcenter[ .blockquote[ ### Example: War in Ukraine and Oil prices - The global market price of crude oil was ~$76 per barrel in January 2022..fn[2] The war drove prices to $110 at around March 2022 and increased volatility substantially &rarr; an *unstable* DGP. - Data mining methods tend to make bad forecasts given such DGP change and also fail to provide reliable measures of forecast uncertainty. - Models based on economic and statistical theory better adapt to new (market) conditions. ]] .footnote[[2] see https://www.gep.com/blog/mind/russia-ukraine-wars-effects-oil-and-gas-industry#:~:text=Oil%20prices%20were%20rising%20globally,barrel%20on%204%20March%202022.] --- # Data mining &mdash; looking for patterns .vcenter[ .blockquote[ ### Example: Viagra **To find a pattern can make you rich** - Pfizer scientists originally developed Viagra scientists to mitigate high blood pressure and angina. - Famous side effects have been found using data mining: instead of coming to the data with a theory, the scientists noticed an interesting pattern in the data. **Data mining is bad as a final step** - Taking a pattern found as given would be problematic - Pfizer did clinical studies to *see if the pattern replicated*: they verified an hypothesis derived from their research question using experimental data. ]] --- # So what makes a research question a *good* one? **Simplicity** - Do not bundle a bunch of research questions into one. - Consider potential answers &mdash; research question and theory probably do not align very well if we cannot place potential answers within the framework of our theory. **Feasibility / scalability ** - A good research question can be answered using the right data. - Data acquisition is often subject to constraints &mdash; think time, money, data protection laws. **Suitable research design** - The research design must be suitable to investigate the derived hypothesis. - Thoughtful consideration must be given as to which statistical method is appropriate. We will deal with this aspect in particular in this course.